Multi-agent security cooperative control method, system and electronic equipment based on risk classification event triggering and topology self-adaptive adjustment
By employing risk-based event triggering and topology adaptive adjustment, the problems of communication resource waste and topology adaptability in multi-agent systems in complex environments are solved, thereby improving security and collaborative performance.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- HUAIYIN INSTITUTE OF TECHNOLOGY
- Filing Date
- 2026-04-22
- Publication Date
- 2026-07-10
AI Technical Summary
Existing multi-agent systems suffer from problems such as wasted communication resources, difficulty in adapting fixed topologies to environmental changes, and insufficient coupling between cooperative control and security control in complex dynamic environments.
A method based on risk-level event triggering and topology adaptive adjustment is adopted. Risk is classified by constructing a comprehensive risk index, and the communication frequency and local topology are dynamically adjusted. Combined with security constraints, the collaborative control input is optimized.
It enables the reduction of communication resource waste, improvement of critical information exchange efficiency, and ensures system security and collaborative performance in complex environments, thereby enhancing the reliability of task execution.
Smart Images
Figure CN122363331A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of multi-agent cooperative control and intelligent control technology, and in particular to a multi-agent safe cooperative control method, system and electronic device based on risk-level event triggering and topology adaptive adjustment, which can be applied to safe cooperative control scenarios of UAV swarms, mobile robot swarms, unmanned vehicle fleets, unmanned surface vessel formations and other distributed autonomous systems. Background Technology
[0002] Multi-agent systems, due to their advantages such as distributed decision-making, self-organization and coordination, parallel task execution, and strong robustness, are widely used in mission scenarios such as formation flying, cooperative inspection, area coverage, target acquisition, warehousing and transportation, and disaster search and rescue. To address the cooperative control problem of multi-agent systems, existing technologies have proposed various methods, including consensus control, formation control, distributed optimization control, model predictive control, and event-triggered control.
[0003] However, in complex and dynamic environments, existing technologies still have the following shortcomings:
[0004] (1) Many existing methods use periodic sampling and fixed communication mechanisms. Each agent needs to continuously broadcast state information within a fixed period. Even when the system state is stable or the environmental risk is low, a large amount of redundant communication will still be generated, resulting in high bandwidth consumption, high energy consumption and low resource utilization.
[0005] (2) Most existing event triggering methods only set a single fixed triggering threshold based on state error, which cannot comprehensively reflect the actual operational risks of multi-agent systems. When an agent approaches an obstacle, the distance between neighbors decreases, or the local communication link deteriorates, the single error triggering mechanism is difficult to increase the update frequency in a timely manner; while in low-risk stages, it may cause unnecessary frequent triggering.
[0006] (3) Existing multi-agent cooperative control methods usually use a pre-given fixed topology, which makes it difficult to adjust the adjacency relationship in real time according to environmental changes, local risk states and communication quality. When local areas are congested, blocked by obstacles, have abnormal nodes or link attenuation, the fixed topology may lead to insufficient exchange of key information, reducing the cooperative performance and security of the system.
[0007] (4) In complex environments, multi-agent systems need to simultaneously satisfy group collaborative goals and individual safety constraints. In existing technologies, collaborative control and safety control are often designed separately, lacking a unified coordination mechanism. This can easily lead to problems such as prioritizing collaborative goals resulting in insufficient safety margins, or excessive safety avoidance leading to a decline in overall task performance.
[0008] Therefore, there is an urgent need for a multi-agent safety collaborative control method that can dynamically adjust the communication trigger frequency based on system operation risks, adjust the local topology in real time based on risk status, and integrate safety constraints and collaborative control. Summary of the Invention
[0009] Purpose of the invention: This invention provides a multi-agent safety collaborative control method, system, and electronic device based on risk-level event triggering and topology adaptive adjustment, to solve the problems of insufficient risk state perception, difficulty in adapting fixed topology to dynamic environmental changes, and insufficient coupling between collaborative control and safety control in the prior art.
[0010] Technical Solution: This invention discloses a multi-agent safety cooperative control method based on risk-level event triggering and topology adaptive adjustment, comprising the following steps:
[0011] Establish dynamic models, communication topology models, and environmental models for multi-agent systems;
[0012] For each intelligent agent, a comprehensive risk index is constructed based on its relative distance to neighboring intelligent agents, distance to obstacles, communication link quality, and local coordination error. The risk level of each intelligent agent's current operating status is then classified according to the comprehensive risk index to obtain the corresponding risk level.
[0013] Based on the risk level and event trigger threshold, perform status broadcasting and control input updates;
[0014] During state broadcasting and control input updates, the communication topology of each agent is adaptively adjusted according to the risk level and the importance of neighbors, the neighbor set and edge weights are updated, and the cooperative control input of each agent is calculated based on the adjusted communication topology and the most recent broadcast state.
[0015] Furthermore, comprehensive risk indicators Collision risk item from neighbors Obstacle risk items Communication quality risk items and cooperative error risk items Weighted composition; Neighbor collision risk item Determined based on the relative distance between neighboring agents, the minimum safe distance between neighboring agents, and the warning distance of neighboring agents; obstacle risk items. The relationship between the minimum distance from the agent to the obstacle boundary, the minimum safe distance from the obstacle, and the warning distance of neighboring agents is used to determine the communication quality risk items. Determining the 1-average link quality index value; Coordination error risk item. It is determined based on the local coordination error index, which is determined by the local position coordination error and the local velocity coordination error.
[0016] Furthermore, risks are classified into low risk, medium risk, and high risk based on a first risk threshold and a second risk threshold, with the first risk threshold being less than the second risk threshold; event trigger thresholds include: a low-risk state corresponds to the first trigger threshold, a medium-risk state corresponds to the second trigger threshold, and a high-risk state corresponds to the third trigger threshold;
[0017] The event trigger function is:
[0018] ;
[0019] in, Indicates the overall triggering error. Indicates the first The trigger threshold corresponding to each agent under the current risk level. Indicates local position error. Indicates local velocity error. It is a positive number.
[0020] Furthermore, the importance of neighbors Based on the relative distance between agents, link quality, and task relevance, it is constructed and represented as:
[0021] ;
[0022] in, Indicates the first The first agent and the second The relative distance between agents This represents the link quality metric between the two. Indicates task relevance metrics. For positive weight parameters, It is a positive number.
[0023] Furthermore, adaptive adjustment of the communication topology for each agent includes:
[0024] When the agent is in a low-risk state, maintain the basic neighbor set;
[0025] When the agent is in a medium-risk state, candidate neighbors with a neighbor importance not lower than the first importance threshold are introduced into the basic neighbor set;
[0026] When the agent is in a high-risk state, candidate neighbors with a neighbor importance not lower than the second importance threshold are introduced, where the second importance threshold is less than the first importance threshold.
[0027] Furthermore, the border weight update is specifically as follows:
[0028] ;
[0029] in, For the new border rights, For the new basic neighbor set, As a comprehensive risk indicator, , The weights represent the importance and risk level of neighbors.
[0030] Furthermore, the input of the collaborative control term is represented as follows:
[0031]
[0032] in, and These are the position feedback gain and velocity feedback gain, respectively. For the adjusted edge weights, and The first The most recent broadcast of the agent's position and velocity states.
[0033] Furthermore, after obtaining the cooperative control input of each agent, the following steps are performed: under the condition of satisfying safety constraints, the cooperative control input is corrected to obtain the final control input of each agent, and the final control input is applied to the corresponding agent; during the correction, the final control input is obtained by solving the problem of minimizing the control input error.
[0034] This invention also discloses a multi-agent safety cooperative control system based on risk-level event triggering and topology adaptive adjustment, comprising:
[0035] The state awareness module is used to acquire the position, velocity, neighbor broadcast status, obstacle information, and link quality information of each agent;
[0036] The risk assessment module is used to construct comprehensive risk indicators based on the information obtained by the status awareness module.
[0037] The risk grading module is used to determine the risk level based on comprehensive risk indicators;
[0038] The event-triggered decision module is used to determine whether to perform status broadcasting and control updates based on the risk level and event trigger presets.
[0039] The topology adjustment module is used to update the neighbor set and edge weights based on the risk level and the importance of neighbors;
[0040] The collaborative control module is used to calculate collaborative control inputs based on the adjusted topology;
[0041] The safety correction module is used to correct the cooperative control input under safety constraints;
[0042] The execution module is used to control the movement of each intelligent agent based on the final control input.
[0043] The present invention also discloses an electronic device, including a processor and a memory, wherein the memory stores a computer program, and when the processor executes the computer program, it implements the above-described method steps.
[0044] Beneficial effects:
[0045] (1) By constructing a comprehensive risk assessment model, collision risk, obstacle threat, communication link status and coordination error are integrated to enable the control system to have risk perception capability.
[0046] (2) By designing a risk-level event triggering mechanism, the status broadcast frequency and control update frequency can be adaptively changed with the risk level, reducing redundant communication in the low-risk stage and improving response speed in the high-risk stage.
[0047] (3) By introducing a topology adaptive adjustment mechanism, the multi-agent system can dynamically adjust the topology connection based on local risk and neighbor importance, thereby improving the efficiency of key information interaction in complex environments.
[0048] (4) By embedding safety constraints into the collaborative control input solution process, the unified coordination of group collaborative goals and individual safety constraints is achieved, thereby reducing collision risk and improving the reliability of task execution.
[0049] (5) The present invention has good scenario versatility and engineering feasibility, and can be widely applied to various multi-agent platforms such as drones, mobile intelligent agents, unmanned vehicles and unmanned boats. Attached Figure Description
[0050] Figure 1 This is a flowchart of the overall process of the method of the present invention. Figure 2 This is a schematic diagram of a multi-agent system scenario. Figure 3 This is a structural diagram of the risk assessment and risk classification module. Figure 4 Trigger risk curves and average / maximum risk graphs for each agent.
[0051] Figure 5 This is a comparison chart of the distances between the smallest intelligent agents. Figure 6 This is a comparison chart of the number of communications under the fixed-period communication and fixed-topology methods and the method of this invention.
[0052] Figure 7 This is a comparison chart of recovery time under fixed-period communication and fixed-topology methods with the method of this invention.
[0053] Figure 8 This is a schematic diagram of the trajectory of multiple agents traversing an obstacle area in the embodiment. Detailed Implementation
[0054] The present invention will be further described below with reference to the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and should not be used to limit the scope of protection of the present invention.
[0055] Example 1:
[0056] This invention discloses a multi-agent safety cooperative control method based on risk-level event triggering and topology adaptive adjustment, comprising the following steps:
[0057] Step 1: Establish the dynamics model, communication topology model, and environment model of the multi-agent system.
[0058] Assume a multi-agent system consists of It consists of 3 intelligent agents, denoted as the 1st agent. The position vectors of the agents are The velocity vector is The control input is ,in, The dimension is either two-dimensional or three-dimensional.
[0059] In one implementation, each agent satisfies a second-order integrator model:
[0060]
[0061] Define the time-varying communication graph as follows:
[0062]
[0063] in, For a set of nodes, Let be the set of edges. Let be an adjacency matrix. If the ... The first agent and the second An intelligent agent at time If communication is possible, then there is ,otherwise .
[0064] No. The neighbor set of an agent is defined as:
[0065]
[0066] The Graph Laplacian matrix is defined as:
[0067]
[0068] in, It is a degree matrix.
[0069] The system's collaborative goal is:
[0070]
[0071] in, Let || be the desired relative displacement vector, with magnitude || || represents the expected relative distance.
[0072] Definition of cooperative error:
[0073] For the For each intelligent agent, the local position cooperative error and the local velocity cooperative error are defined as follows:
[0074]
[0075] The local cooperative error index is defined as follows:
[0076] in, , .
[0077] Step 2: For each agent, construct a comprehensive risk index based on its relative distance to neighboring agents, distance to obstacles, communication link quality, and local coordination error.
[0078] Define the comprehensive risk function as follows:
[0079]
[0080] in: For neighbor collision risk items; For obstacle risk items; This is a communication quality risk item; For the risk item of cooperative error, , .
[0081] Neighbor collision risk items
[0082] Let the relative distance be:
[0083]
[0084] Let the minimum safety distance be The warning distance is ,and .definition:
[0085]
[0086] but:
[0087]
[0088] Obstacle risk items
[0089] Let the first The minimum distance from each agent to the obstacle boundary is:
[0090]
[0091] in, Indicates the first There are several obstacles. Assume the minimum safe distance between the obstacles is... The warning distance is ,but:
[0092]
[0093] Communication quality risk items
[0094] Let the link quality index be Define average link quality:
[0095]
[0096] The communication quality risk item is:
[0097]
[0098] Cooperative error risk item
[0099] definition:
[0100]
[0101] in, .
[0102] Step 3: Classify the current operating status of each agent according to the comprehensive risk index to obtain the corresponding risk level.
[0103] Set two risk thresholds: First risk threshold Second risk threshold ,satisfy: .
[0104] The risk level is defined as follows:
[0105]
[0106] in , , These represent low-risk, medium-risk, and high-risk states, respectively.
[0107] Step 4: Determine the corresponding event trigger threshold based on the risk level, and determine whether to perform status broadcasting and control input updates based on the triggering conditions. See also Figure 4 Risk curves and average / maximum risk graphs for each agent.
[0108] Risk classification event triggering mechanism
[0109] Let the first The broadcast time of each agent is ,exist The most recent broadcast status is:
[0110]
[0111] Define the sampling error corresponding to the most recent broadcast state as:
[0112]
[0113] Based on the comprehensive risk value, the system status is divided into three levels: low risk, medium risk, and high risk, and a graded trigger threshold is defined:
[0114]
[0115] in, .
[0116] Event triggering function:
[0117] ;
[0118] when Status broadcasts and control updates are triggered periodically.
[0119] Step 5: Based on the risk level and the importance of neighbors, adaptively adjust the communication topology of each agent and update the neighbor set and edge weights.
[0120] Topology adaptive adjustment mechanism
[0121] The importance of neighbors is defined as follows:
[0122]
[0123] in: It is a relative distance; For link quality; For task relevance indicators; , It is a positive number.
[0124] If the agent is in a low-risk state, then maintain the basic neighbor set:
[0125]
[0126] If the risk level is medium, then:
[0127]
[0128] If the situation is high-risk, then:
[0129]
[0130] in .
[0131] The edge weights are updated as follows:
[0132] .
[0133] in, For the new border rights, For the new basic neighbor set, As a comprehensive risk indicator, , The weights represent the importance and risk level of neighbors.
[0134] Step 6: Based on the adjusted communication topology and the most recent broadcast state, calculate the cooperative control input for each agent.
[0135] Step 7: Under the condition of satisfying safety constraints, modify the cooperative control input to obtain the final control input of each agent.
[0136] Step 8: Apply the final control input to the corresponding agent to achieve safe and collaborative control of the multi-agent system.
[0137] Cooperative control laws and safety corrections
[0138] Define the control input as:
[0139]
[0140] in, Indicates a safety correction item. For collaborative control items:
[0141] To ensure safety among intelligent agents, a safety function (safety constraint) is defined:
[0142]
[0143] Require:
[0144]
[0145] Its derivative is:
[0146]
[0147] Further, there are:
[0148]
[0149] Apply higher-order barrier constraints:
[0150]
[0151] Define a safety function for each obstacle:
[0152]
[0153] Constraints are applied only to obstacles within the perception range:
[0154]
[0155] If the obstacle is circular, the center... ,radius It can be written directly as:
[0156]
[0157] Its derivative is:
[0158]
[0159] The second derivative is:
[0160]
[0161] Apply constraints:
[0162] The final control input is solved through the following optimization:
[0163]
[0164] The above-mentioned safety constraints between intelligent agents and obstacles are satisfied.
[0165] Stability analysis
[0166] Within any fixed topology interval after topology adjustment, the closed-loop stability of the system can be analyzed as follows, and the Lyapunov function can be constructed:
[0167]
[0168] Differentiating it, we get:
[0169]
[0170] Using Young's inequality:
[0171]
[0172] Then we have:
[0173]
[0174] if only
[0175] in, and These represent the overall position error vector and the overall velocity error vector, respectively; for 3D identity matrix The dimension of the space; , These represent the position feedback gain and the velocity feedback gain, respectively. This represents the aggregated disturbance term introduced by event-triggered sampling error and broadcast state error; These are positive constants introduced when applying Young's inequality; Represents the Laplace matrix The second smallest eigenvalue.
[0176] Therefore, under the constraint of event-triggered error, the system error is uniformly and eventually bounded. This is true if the system state and nominal control input are bounded, i.e., if there exists a constant. , making
[0177]
[0178] The rate of increase of the trigger error is bounded, that is, there exists a constant. , so that:
[0179]
[0180] Because after each trigger:
[0181]
[0182] The next trigger requires at least reaching the threshold. Therefore, there is a positive lower bound between the two triggers:
[0183]
[0184] Therefore, the Zeno phenomenon does not exist in the system.
[0185] Meanwhile, since the safety barrier constraints are always satisfied, the safety set remains forward invariant to the closed-loop system, thus ensuring that the safety distance constraints between agents and between agents and obstacles are always valid.
[0186] In one exemplary embodiment, using a formation of six two-dimensional mobile intelligent agents for obstacle avoidance as an example, see [link to example]. Figure 2 The dynamics of each intelligent agent satisfy:
[0187]
[0188] The parameter settings are as follows:
[0189]
[0190] The system is initially in a low-risk state with a low communication frequency. When the agent approaches an obstacle area or the distance between neighbors decreases, the overall risk increases, the trigger threshold decreases, the communication update frequency increases, and the local topology is enhanced. Under the influence of safety constraints, the system completes obstacle avoidance and restores the target formation.
[0191] Simulation results show that, compared with fixed-period communication and fixed-topology methods, the multi-agent safety cooperative control method based on risk-level event triggering and topology adaptive adjustment proposed in this invention can significantly reduce the number of communications and improve the formation recovery efficiency after obstacle areas, while satisfying the minimum safe distance constraint. For details, see... Figure 5 and Figure 6 In this embodiment, under the fixed-period communication and fixed-topology method, each agent broadcasts its status and updates its control according to a preset fixed period, and the adjacency relationship remains unchanged. In the method of the present invention, each agent constructs a comprehensive risk index based on the relative distance of neighbors, the distance of obstacles, the quality of communication links and local cooperative errors, and dynamically adjusts the event triggering threshold and local topology connection relationship according to the comprehensive risk state, thereby realizing the adaptive adjustment of communication update frequency and communication topology.
[0192] like Figure 6 As shown, under the fixed-period communication and fixed-topology method, the system maintains a high communication frequency throughout the entire task execution process, resulting in a high number of communications for each agent and a high total number of system communications. However, by adopting the method of this invention, the state broadcast frequency is reduced by using a higher trigger threshold during low-risk phases, and the information interaction between key nodes is enhanced by reducing the event trigger threshold and combining it with local topology adjustments during medium- and high-risk phases. This allows communication resources to be more concentrated in the critical phases where system state changes are significant. In this embodiment, the total number of system communications under the fixed-period communication and fixed-topology method is 2154, while the total number of system communications after adopting the method of this invention is 124, demonstrating that this invention can significantly reduce the number of communications and lower the overall communication burden of the system.
[0193] Furthermore, such as Figure 7As shown, in the formation recovery phase after the obstacle zone, the recovery time under the fixed-period communication and fixed-topology methods is 8.25 s, while the recovery time using the method of this invention is 8.15 s. This result demonstrates that this invention does not simply reduce the communication frequency, but rather prioritizes event-triggered communication updates and local topology adjustments in critical operational phases based on the comprehensive risk state. This allows each agent to achieve more effective information interaction during local formation adjustments and overall formation recovery after obstacle avoidance, thereby improving the formation recovery efficiency after the obstacle zone. See also... Figure 8 This is a schematic diagram of the trajectory of multiple agents traversing the obstacle area in this embodiment.
[0194] In summary, the multi-agent safety cooperative control method based on risk-level event triggering and topology adaptive adjustment proposed in this invention can dynamically adjust the communication update frequency and optimize the local communication topology in real time while meeting safety constraints. This effectively reduces the number of communications, lowers redundant communication overhead, and improves the formation recovery efficiency after obstacle areas in multi-agent formation obstacle avoidance tasks. This verifies that the method of this invention can balance the reduction of communication overhead and the improvement of formation recovery performance, and has good engineering application value.
[0195] Example 2:
[0196] This invention discloses a multi-agent safety cooperative control system based on risk-level event triggering and topology adaptive adjustment, comprising:
[0197] The state awareness module is used to acquire the position, velocity, neighbor broadcast status, obstacle information, and link quality information of each agent.
[0198] The risk assessment module is used to construct comprehensive risk indicators based on the information obtained from the status awareness module.
[0199] The risk grading module is used to determine the risk level based on comprehensive risk indicators.
[0200] The event-triggered decision module is used to determine whether to perform status broadcasting and control updates based on the risk level and event trigger presets.
[0201] The topology adjustment module is used to update the neighbor set and edge weights based on the risk level and the importance of the neighbors.
[0202] The collaborative control module is used to calculate collaborative control inputs based on the adjusted topology.
[0203] The safety correction module is used to correct cooperative control inputs under safety constraints.
[0204] The execution module is used to control the movement of each intelligent agent based on the final control input.
[0205] The above modules work together to implement the multi-agent safety collaborative control method based on risk-level event triggering and topology adaptive adjustment as described in Example 1.
[0206] Example 3:
[0207] The present invention also discloses an electronic device, including a processor and a memory, wherein the memory stores a computer program, and when the processor executes the computer program, it implements the method of multi-agent safe cooperative control based on risk-level event triggering and topology adaptive adjustment in Embodiment 1 above.
[0208] The above embodiments are only for illustrating the technical concept and features of the present invention, and are intended to enable those skilled in the art to understand the content of the present invention and implement it accordingly. They should not be construed as limiting the scope of protection of the present invention. All equivalent transformations or modifications made in accordance with the spirit and essence of the present invention should be covered within the scope of protection of the present invention.
Claims
1. A multi-agent safety cooperative control method based on risk-level event triggering and topology adaptive adjustment, characterized in that, Includes the following steps: Establish dynamic models, communication topology models, and environmental models for multi-agent systems; For each intelligent agent, a comprehensive risk index is constructed based on its relative distance to neighboring intelligent agents, distance to obstacles, communication link quality, and local coordination error. The risk level of each intelligent agent's current operating status is then classified according to the comprehensive risk index to obtain the corresponding risk level. Based on the risk level and event trigger threshold, perform status broadcasting and control input updates; During state broadcasting and control input updates, the communication topology of each agent is adaptively adjusted according to the risk level and the importance of neighbors, the neighbor set and edge weights are updated, and the cooperative control input of each agent is calculated based on the adjusted communication topology and the most recent broadcast state.
2. A multi-agent safety cooperative control method based on risk-level event triggering and topology adaptive adjustment according to claim 1, characterized in that, Comprehensive risk indicators Collision risk item from neighbors Obstacle risk items Communication quality risk items and cooperative error risk items Weighted composition; Neighbor collision risk item Determined based on the relative distance between neighboring agents, the minimum safe distance between neighboring agents, and the warning distance of neighboring agents; Obstacle risk items The relationship between the minimum distance from the agent to the obstacle boundary, the minimum safe distance from the obstacle, and the warning distance of neighboring agents is used to determine the communication quality risk items. Determining the 1-average link quality index value; Coordination error risk item. It is determined based on the local coordination error index, which is determined by the local position coordination error and the local velocity coordination error.
3. A multi-agent safety cooperative control method based on risk-level event triggering and topology adaptive adjustment according to claim 1, characterized in that, Risk is classified into low risk, medium risk and high risk based on the first risk threshold and the second risk threshold, with the first risk threshold being less than the second risk threshold. Event trigger thresholds include: a first trigger threshold for low-risk states, a second trigger threshold for medium-risk states, and a third trigger threshold for high-risk states. The event trigger function is: ； in, Indicates the overall triggering error. Indicates the first The trigger threshold corresponding to each agent under the current risk level. Indicates local position error. Indicates local velocity error. It is a positive number.
4. A multi-agent safety cooperative control method based on risk-level event triggering and topology adaptive adjustment according to claim 1, characterized in that, Neighbor importance Based on the relative distance between agents, link quality, and task relevance, it is constructed and represented as: ； in, Indicates the first The first agent and the second The relative distance between agents This represents the link quality metric between the two. Indicates task relevance metrics. For positive weight parameters, It is a positive number.
5. A multi-agent safety cooperative control method based on risk-level event triggering and topology adaptive adjustment according to claim 4, characterized in that, Adaptive adjustment of the communication topology for each agent includes: When the agent is in a low-risk state, maintain the basic neighbor set; When the agent is in a medium-risk state, candidate neighbors with a neighbor importance not lower than the first importance threshold are introduced into the basic neighbor set; When the agent is in a high-risk state, candidate neighbors with a neighbor importance not lower than the second importance threshold are introduced, where the second importance threshold is less than the first importance threshold.
6. A multi-agent safety cooperative control method based on risk-level event triggering and topology adaptive adjustment according to claim 5, characterized in that, The specific steps for updating edge weights are as follows: ； in, For the new border rights, For the new basic neighbor set, As a comprehensive risk indicator, , The weights represent the importance and risk level of neighbors.
7. A multi-agent safety cooperative control method based on risk-level event triggering and topology adaptive adjustment according to claim 6, characterized in that, The input for the coordinated control item is represented as follows: in, and These are the position feedback gain and velocity feedback gain, respectively. For the adjusted edge weights, and The first The most recent broadcast of the agent's position and velocity states.
8. A multi-agent safety cooperative control method based on risk-level event triggering and topology adaptive adjustment according to claim 1, characterized in that, After obtaining the cooperative control input of each agent, the following steps are performed: under the condition of satisfying safety constraints, the cooperative control input is corrected to obtain the final control input of each agent, and the final control input is applied to the corresponding agent; During the correction, the final control input is obtained by minimizing the control input error.
9. A multi-agent safety cooperative control system based on risk-level event triggering and topology adaptive adjustment, characterized in that, include: The state awareness module is used to acquire the position, velocity, neighbor broadcast status, obstacle information, and link quality information of each agent; The risk assessment module is used to construct comprehensive risk indicators based on the information obtained by the status awareness module. The risk grading module is used to determine the risk level based on comprehensive risk indicators; The event-triggered decision module is used to determine whether to perform status broadcasting and control updates based on the risk level and event trigger presets. The topology adjustment module is used to update the neighbor set and edge weights based on the risk level and the importance of neighbors; The collaborative control module is used to calculate collaborative control inputs based on the adjusted topology; The safety correction module is used to correct the cooperative control input under safety constraints; The execution module is used to control the movement of each intelligent agent based on the final control input.
10. An electronic device, characterized in that, It includes a processor and a memory, the memory storing a computer program, and when the processor executes the computer program, it implements the method steps of any one of claims 1 to 8.